Traditional broadcast audience measurement systems utilize single-source data panels consisting of individuals or households recording media exposure over time. Because these individuals are typically paid for participation, measurement data tends to be of high quality, but for a limited population size. Accordingly, statistical confidence in applying the measurement data to a large population may be low. Conversely, online audience measurement systems may obtain measurement data from a very large population, but this data may be of questionable quality, as users may provide false or incomplete demographic data for various reasons.